Patient Safety Tip of the Week


January 6, 2009         Preventing Inpatient Suicides






It has been almost 20 years since New York State issued a report on preventing inpatient suicides and over 10 years since Joint Commission first published a Sentinel Event Alert on prevention of inpatient suicides. Though many excellent recommendations have been made, the problem continues to occur. Inpatient suicide still accounts for 12.4% of all sentinel events collected by Joint Commission over the years. The hospital incident reporting system in New York (NYPORTS) had 278 reported attempted suicides in the 3-year period from 2002-2004.


The original New York report focused on hangings because that was the most common mode of suicide in inpatients. Most of these hangings occurred in relatively secluded areas, especially bathrooms, closets and private or semiprivate bedrooms. Many of these occurred despite patients being on 15-minute observation status. They identified many environmental factors and procedural factors that might be modified to help prevent inpatient suicides.


Environmental factors included exposed pipes, showerheads, bars in showers and toilet stalls. The commission recommended recommended environmental safeguards such as use of breakaway physical structures. Joint Commission also recommends that breakaway hardware actually be tested regularly (weight testing to ensure that hardware will, in fact, break away when sufficient weight is applied). Other environmental safeguards include keeping doors closed, and keeping sharps or other hazardous materials away from patients. The Joint Commission Sentinel Event Alert noted 75% of inpatient suicides were by hanging but also noted that 20% involved jumping from roofs or windows. So careful attention to access to such sites is important.


Procedural factors were also important. They noted that physician-ordered suicide observation status was in effect in about a third of all the patients at the time of suicide but that in half the cases those observations were not being carried out appropriately. The manner in which the physician wrote the order was important. Explicit orders were more likely to be carried out than were vaguely-worded orders.


Obviously, patient-related factors are important and patients with depression and affective disorders are at greatest risk. Patients with multiple diagnoses (eg. depression and substance abuse) are at higher risk for suicide but there may also be different risk profiles for different demographics as well.


Though the original New York report focused on psychiatric facilities, many inpatient suicides occur on med-surg units and units other than inpatient psychiatric units. The Joint Commission Sentinel Event Alert noted that of 27 suicides occurring in general hospitals, 12 occurred in med-surg units and one in the emergency room. An AHRQ WebM&M Case & Commentary several years ago highlighted many of the issues involved in cases where patients are admitted to non-psychiatric services. There are a variety of reasons why suicides may occur on non-psychiatric services. While most psychiatric units now have rigorous policies and procedures about assessing the environment for hazards that could be used for suicide, the same rigor is seldom applied on med-surg units. In the AHRQ case noted above, the suicide attempt occurred in a bathroom in the radiology suite. We wonder how many hospitals would have inspected that bathroom for potential suicide risk. None are likely to have done such inspection as part of their routine environmental safety process. So it would have to be done ad hoc at a time a potentially suicidal patient is brought to the area. We’ve talked about the “Ticket to Ride” communication tool for hospital transports. Perhaps even that useful tool needs to be modified to account for potentially suicidal patients.


Suicide risk assessments are often incomplete or not done and the Joint Commission especially noted a dearth of suicide reassessments. Sound familiar? How often have we noted that other key assessments in healthcare, such as fall risk or DVT risk assessment, are done on admission but are not repeated even though clinical circumstances have changed during the hospital course? Another issue is that even though the reason for admission may have been related to a suicide attempt, the patient on admission may be unable to cooperate with a suicide risk assessment (eg. the patient may be comatose or obtunded because of a drug overdose). Sometimes many days pass where the patient physically would have been incapable of another suicide attempt but, ironically, as they begin to improve medically the suicide risk reappears.


As in most sentinel events, communication issues are often root causes. The Joint Commission emphasized issues such as training/education, not only of staff but also of family and friends. An Austrailian study on inpatient suicides specifically mentions the risk of suicide in patients on leave or pass and several other studies have also noted that association.


An excellent review, based on root cause analyses (RCA’s) of inpatient suicides and suicide attempts in the VA Hospital system, appears in the August 2008 Joint Commission Journal on Quality and Patient Safety (Mills et al 2008). About half their events occurred on non-psychiatric units and the methods differed by site. Whereas hanging/asphyxiation, cutting, and fires occurred most often on the psychiatric units, overdoses, jumpings, stabbings, and ingestion of chemicals was more common on non-psychiatric units. They provide tables describing the types of anchor points in hangings, the materials used as nooses, the implements used in cuttings, and the loctions for jumpings. They point out that, for a variety of reasons, it may be impossible to eliminate the materials used for nooses so they suggest a focus on eliminating anchor points. For example, they note that interior doors and cabinets can be removed or replaced with accordion doors that cannot be used as anchor points. And they recommend that things like door knobs, railings, faucets and hooks be eliminated or constructed so as to break away when weight is applied. They also note that most of the attempted overdoses took place on units other than psychiatric inpatient units (though many of these units were detox units, etc.) so careful assessment of security of medications is important on all units. They also note that the VA has developed a mental health environment-of-care checklist that is available by e-mail request.


Bathrooms on non-psychiatric floors may be especially problem prone. Not only are they seldom assessed for tools and implements that could be used for suicide, but some also allow the door to be locked from the inside. So observation protocols for potentially suicidal patients on such units should ensure that doors are not locked (or, if they can be locked, that the “observer” has keys to access the bathroom). Having observers of the same gender as the patient also is recommended.


Several articles have noted that the “sitters” commonly utilized to monitor the potentially suicidal patient on the non-psychiatric unit are often not specifically trained in assessment of the environment or management of the suicidal patient. The AHRQ article noted above has a good discussion on this.


Most healthcare organizations that have developed standardized order sets, whether they are paper-based or CPOE-based, tend to focus on the more commonly seen diagnoses and conditions. We’ve previously made the case that standardized order sets may be even more important for conditions or circumstances that are high risk but less likely to be encountered. For example, we have noted that some drugs (eg. argotroban or desmopressin) may need to be used under rare circumstances and that few practitioners have extensive experience with those drug. Therefore, some organizations have made special attempts to make appropriate information on the less familiar drugs available to practitioners and have developed standardized order sets for dealing with such drugs. Using similar logic, it would be appropriate to develop standardized order sets for dealing with the potentially suicidal patient admitted to non-psychiatric services.


Since it is very difficult to predict suicides, careful attention to environmental factors that could facilitate suicide is critical. Appropriate assessment and reassessment for suicide risk are important. Similarly, for those patients identified as at-risk for suicide, it is essential that appropriate monitoring with appropriately trained staff be used. Transitions of care are especially vulnerable and suicide risk should be considered in all handoffs and other communications.




Update: See also our February 9, 2010 Patient Safety Tip of the Week “More on Preventing Inpatient Suicides” and our December 2010 What’s New in the Patient Safety World column “Joint Commission Sentinel Event Alert on Suicide Risk Outside Psych Units”.







New York State Commission on Quality of Care. Preventing Inpatient Suicides: An Analysis of 84 Suicides by Hanging in New York State Psychiatric Facilities (1980-1985).



Joint Commission. Sentinel Event Alert Issue #7. Inpatient Suicides: Recommendations for Prevention. November 6, 1998



Joint Commission. Sentinel Event Statistics as of September 2008.



NYSDOH. NYPORTS Annual Report 2002-2004.



Gibson J, Taylor DH. AHRQ Web M&M Case & Commentary “The Dangerous Detour.” June 2003



Victorian Government Health Information (Australia). Sentinel Event Program. Annual Report 2007-2008.



Mills PD, DeRosier JM, Ballot BA, Sheperd M, Bagian JP. Inpatient Suicide and Suicide Attempts in Veterans Affairs Hospitals. The Joint Commission Journal on Quality and Patient Safety 2008; 34: 482-488 (August 2008)











Patient Safety Tip of the Week Archive


What’s New in the Patient Safety World Archive